Robust hashing learning via multi-view subspace learning

Hashing learning has attracted increasing attention these years with the explosive increase of data. The hashing learning can be divided into two steps. Firstly, obtain the low dimensional representation of the original data. Secondly, quantize the real number vector of the low dimensional representation of each data point and map them to binary codes. Most of the existing methods measure the original data only from one perspective. This paper introduces the multi-view methods to the hashing learning field, and proposes a hashing learning framework utilizing the multi-view methods. The experimental results illustrate that our algorithm outperforms several the other state-of-the-art methods.

I. Introduction

With the rapid development of mobile Internet, social network and cloud computing in recent years, the volume of data increased rapidly. To solve the problem of storage, analysis and management of these data, hashing learning has attracted many researchers attention. Hashing learning projects the original data into binary codes which can reduce the consumption of storage and computing and improve the efficiency of system. Hashing learning has been widely applied on information retrieval [1], [2], data mining [3], [4], pattern recognition [5], [6], multimedia information processing [7], [8] and computer vision [9], [10].